Interactions

When two or more items on your timeline interact, Lab labels the combination. Some interactions are dangerous; some are productive. Lab treats both as first-class — discovering a beneficial pairing is as important as flagging a risky one.

What an interaction is

An interaction is a rule that fires when a specific combination of timeline items is present together (or in a specific timing relationship). The simulation itself never changes — the engine always sees the full timeline — but the rule annotates what's happening so you can read it.

Rules are tag-based, not brand-name-based. "SSRI + MAOI" matches every SSRI in the library, not a hand-curated list. Tags are either declared manually (SSRI, levothyroxine_chelator) or derived from the agent's pharmacology (CYP3A4_substrate_narrow).

Severities

Each rule carries one of four severities, color-coded everywhere it appears:

Severity	Color	What it means
Critical	Red	Hard contraindication. Don't co-administer. Example: serotonergic + MAOI.
Warning	Amber	Real risk worth knowing. Example: GLP-1 + sulfonylurea hypoglycemia.
Benefit	Green	Productive synergy. Example: creatine + carbs (improved uptake), magnesium pre-sleep.
Info	Blue	Educational. Useful context with no specific action.

Positive-valence pairings are not a footnote — they're a primary use of the system. The "Find more synergy" affordance in the inspector helps you discover beneficial combinations you haven't added yet.

Categories

Severity tells you how much it matters. Category tells you what kind of interaction it is:

• drug-drug — two compounds interacting
• drug-food — a compound's PK or PD changes with food
• drug-condition — a compound interacts with a flagged condition on your profile
• absorption-timing — chelation, uptake-window, or empty-stomach rule
• lifestyle-stack — sleep + meditation + light interventions reinforcing each other
• timing-synergy — order or proximity matters (creatine post-workout, magnesium 60 min before sleep)
• nutrient-pairing — fat-soluble vitamin + dietary fat, iron + vitamin C

Where interactions show up in the UI

• Timeline item border — affected items get a colored left border matching the worst-severity rule firing on them. A small chip on the item shows the count.
• Inspector → Interactions section — when you select a single item with one or more interactions, a section near the top of the inspector lists every rule firing on it, sorted by severity, with one-line mechanisms.
• Right-drawer Inspector tab badge — if the drawer is on a non-Inspector tab, a count badge appears on the Inspector tab so you don't miss it.
• Chart pill — when a rule names affected signals, a small pill appears on those signal lines at the windows the rule is active.
• Sim tab — the AI assistant sees your active interactions in its context and will reference them when proposing changes.

Interaction detail view

Click any row in the Interactions section to open the detail view. It shows:

• Mechanism + long explanation — what's happening biologically
• What to do — remediation prose for warnings and criticals
• Items involved — chips for the timeline items that triggered the rule
• Affected signals — clickable chips that jump to the signal in the inspector
• Citations — every paper anchoring the rule, with hover-to-preview and click-to-open
• Derivation — Clinical look-up, Model-emergent, or Mechanistic inference
• PharmD review status — every rule is reviewed by a contracted PharmD before it leaves pre-release. Pending rules carry a clear "PharmD review pending" chip.

Evidence requirements

Interactions don't ship without sources. The citation coverage gate enforces a minimum per severity:

• Critical — at least one high-confidence source (CPIC guideline, FDA label, meta-analysis, or systematic review)
• Warning / Benefit — at least one high OR two medium-confidence sources
• Info — at least one source of any confidence

If a rule's pharmdPending flag hasn't been cleared, the UI surfaces it; the same flag is a CI violation in the citation coverage report.

Dismissing an info-level rule

Some info-level rules are persistently relevant only the first time you see them ("creatine loading phase explained"). The detail view's Dismiss button hides that rule for the current scenario. Critical and warning rules cannot be dismissed.

Why interactions are a labeling layer, not a sim modifier

Conceptually, every interaction is already in the simulation — the underlying drugs, foods, and lifestyle items contribute to the same engine state regardless of whether a rule fires. Rules exist to name what's happening so you can read the chart with intent. Adding or removing a rule never changes a single point on the chart; it only changes what's labeled.

This is why the same combination can fire multiple rules at once — different aspects of the same biology each get their own annotation.

See also

• Citation pills — the same color language used on every cited claim
• The inspector — where the Interactions section lives